>Another idea is to just apply the current operation on the >x+Tx*i,y+Tx*i position too, and skip the tiled view. This might be >easier. I think it's harder, since all ops have to be replicated that way, including filters etc On the other hand, having brushes work modulo the image size might be handy, but is perhaps orthogonal to having a tiled view mode. Having a tiled view in 100% scale while editing the same image in a magnified window would be very handy when making tiles. But this could be implemented more generally by having a view constructed procedurally by a script --- perhaps trigged by an event mechanism, activated each time the view needs re-rendering. Preferably this should happen in the background to minimize interactive latency for the user. This would give the Gimp some of the what-if capability of spreadsheets. You work in one window as usual, and another window dynamically displays the results of certain transforms applied to the first image. This can be tiling, filters, combinations of other images etc. I can not only see the tile repeated next to itself, but also how it combines with other tiles
